Messages in this thread | | | From | Keith Owens <> | Subject | Re: [PATCH] 2.4.20 unresolved symbols with K7/3DNOW | Date | Sun, 23 Mar 2003 15:10:57 +1100 |
| |
On Sun, 23 Mar 2003 04:49:35 +0100, "dose" <dose@infernum.com> wrote: >This one seems to have been around for quite a while (2.4.0-textX ?)...when >compiling with CONFIG_MK7=y ("Athlon/Duron/K7" as Processor family) a lot of >modules aren't able to resolve the symbol _mmx_memcpy. This is caused by >CONFIG_X86_USE_3DNOW=y and the problem lies in a wrong symbol export in >arch/i386/kernel/i386_ksyms.c - I wonder why this still hasn't been fixed as >it doesn't seem like a hard thing to do for someone who's into kernel >patching...being new to that, it took me about 3 hours. > >diff -urN linux-2.4.20/arch/i386/kernel/i386_ksyms.c.orig >linux-2.4.20/arch/i386/kernel/i386_ksyms.c >--- linux-2.4.20/arch/i386/kernel/i386_ksyms.c.orig Sun Mar 23 04:24:06 >2003 >+++ linux-2.4.20/arch/i386/kernel/i386_ksyms.c Sun Mar 23 04:24:18 2003 >@@ -120,7 +120,7 @@ > #endif > > #ifdef CONFIG_X86_USE_3DNOW >-EXPORT_SYMBOL(_mmx_memcpy); >+EXPORT_SYMBOL_NOVERS(_mmx_memcpy); > EXPORT_SYMBOL(mmx_clear_page); > EXPORT_SYMBOL(mmx_copy_page); > #endif
2.4 has a broken build system. With CONFIG_MODVERSIONS=y, you must make mrproper when changing any config option that changes the list of exported symbols. Adding _NOVERS is just papering over the cracks in the kernel build system, the only time _NOVERS should be used is when the exported symbol is in ASM code and not in C.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |